The Aave liquidation cascade risk that has been building inside the protocol’s concentrated liquid staking token (LST) loop trade came within striking distance of activation on 20 August 2026, when Ethereum surged roughly 18% from $1,920 to above $2,270 in a single session. The move did not trigger it. That is almost the worst outcome possible.
When ETH rises, health factors improve. The 9% of Aave positions that carry approximately half the protocol’s $12.2 billion TVL in debt, running an average health factor of 1.06 and a debt-to-equity ratio near 10.7 times, were temporarily safer on 20 August. Wrapper tokens rallied in line with spot ETH, pegs held, and the leverage looked fine. The rally also gave traders a reason to add more recursion loops. If the concentrated cohort added positions during the surge, health factors will have reverted to the same 1.06 level at higher dollar values, meaning the absolute amount at risk has grown even though the percentage buffer is unchanged.
What the March Oracle Incident Revealed About the Aave Liquidation Cascade Risk
The most instructive precedent is the 10 March 2026 incident. According to the Aave governance post-mortem published by LlamaRisk, a newly deployed CAPO Risk Agent pushed a parameter update that artificially depressed the wstETH oracle price by approximately 2.85%. The live on-chain exchange rate was approximately 1.228 wstETH per ETH; the oracle applied 1.1939, a gap large enough to push 34 high-leverage E-Mode positions below their liquidation thresholds automatically.
The result: $26.6 million in liquidation volume across Aave V3 Core and Prime, with 10,938 wstETH liquidated across those 34 accounts. As reported by CryptoRank, liquidation bots captured approximately 499 ETH in bonuses (roughly $1.2 million) by executing against positions that should not have been eligible. The protocol realised a loss of 345 ETH, recovering approximately 141.5 ETH through BuilderNet refunds plus around 13 ETH in associated fees.
That incident involved a 2.85% artificial price distortion across 34 accounts. The structural vulnerability being discussed now involves a potential 8–9% organic wrapper discount across hundreds of accounts simultaneously. The scale comparison is not comforting.
Galaxy’s Stress Test: Collateral Falls Below Debt at a 10% Depeg
Galaxy Research’s depeg stress analysis puts concrete numbers on the scenario. If weETH were to trade at a 10% discount to ETH, Aave would be left with approximately $2.47 billion in debt against approximately $2.42 billion in post-shock collateral, collateral dips below debt. weETH alone accounts for roughly 48% of E-Mode posted collateral; combined with rsETH and wstETH, the top three cover approximately 80% of E-Mode collateral.
rsETH presents a second concentration point. At a 10% depeg, Galaxy estimates approximately $1.16 billion in debt against approximately $1.14 billion in post-shock collateral. That entire exposure is held by only 22 accounts.
Across the full Aave book, WETH, weETH, and wstETH together account for roughly 58.7% of enabled collateral, per Galaxy Research. The looping trade is not a niche strategy. It is the dominant collateral structure on the protocol.
Historical context matters here. According to an Aave governance ARFC on ETH and BTC collateral efficiency, wstETH and weETH together account for 6,486 liquidations and roughly $361 million in seized collateral across all Aave deployments over the past two years, $296 million attributable to wstETH and $47 million to weETH. The median oracle processing lag for LST collateral on Ethereum Core is 12 seconds versus 24 seconds for WETH. In a fast-moving depeg, that 12-second window is what separates an orderly liquidation from a queue of bots competing on gas.
Governance Is Moving, But the Positions Exist Now
Aave’s governance is not sitting still. A temp check proposal on the post-rsETH collateral framework proposes cutting weETH’s E-Mode loan-to-value ratio from 93% to an interim 83%, with a longer-term target of 68%, and reducing the liquidation threshold from 95% to 88%. A separate active ARFC on the unified isolated flag in Aave V3.7 E-Modes proposes isolating the rsETH, ezETH/wstETH, and weETH/wstETH E-Mode categories, all currently non-isolated. A Snapshot vote on low-adoption asset deprecation passed on 15 August 2026.
Governance also has an active proposal to raise liquidation fees for WETH and wstETH on Aave V3 Ethereum Core, which would alter the economics for liquidation bots responding to a cascade. Bots that find the liquidation bonus insufficient relative to gas costs will simply not participate, increasing bad debt risk.
The problem is timing. Governance proposals move through community discussion, Snapshot votes, and on-chain execution over days to weeks. A weETH depeg to 10% below ETH would flip hundreds of accounts below their liquidation thresholds in minutes. According to Talos network analysis, WETH and major stablecoin pools were already running at 80–90% utilisation before the March incident and were pushed to sustained 90–100% utilisation during the shock itself. At 82.46% stablecoin utilisation currently, the buffer before withdrawal liquidity becomes constrained is thin.
The 18% ETH rally bought the concentrated positions another day. Watch weETH, wstETH, and rsETH prices on DEX aggregators: any sustained discount above 3% is the early signal; above 8% the Aave liquidation cascade risk stops being theoretical. The governance proposals closing that window are the only structural fix available, and they are running on a different clock to the market.
